README.md

Python 3.11 multiprocessing and multithreading Benchmarking compared to Python 3.10

The included multi-proc_thread.py file generates random float and int nested lists i.e matrices, then does some manipulations with list comprehensions, lambda and map functions. The iteration loop then mutates matrix dimensions adding more elements in every iteration to get things rather heavy.

Download this repo to your local machine then pull these two docker images :

docker pull python:3.11.0-bullseye

docker pull python:3.10.8-bullseye

Open a CMD in the folder where you put the multi-proc_thread.py file and run this command

docker run -it --rm --privileged -v $(pwd):/myapp python:3.10.8-bullseye bash

it immeditely falls into the docker shell , type ls myapp/ to make sure that you are on the right path :

root@xxxxxx:/# ls myapp/
README.md  multi-proc_thread.py 
root@xxxxxx:/#

To get a fair results close peripheral applications and do not mouse click while testing in both cases.

Paste ' python3 myapp/multi-proc_thread.py' to the shell to run the test. It takes about a minute to finish :

root@xxxxxx:/# python3 myapp/multi-proc_thread.py
Matrix size 10000 .... done!
Matrix size 40000 .... done!
.

['py3.10.8', 'NO_thread() -> 10.1622 s', 'Threads(5,) -> 50.6150 s', 'Processes(5,) -> 15.6486 s']
root@xxxxxx:/# 

copy/paste the final output into a text file for later comparison.

Shut down the container by 'CTRL+D` then start the second container

docker run -it --rm --privileged -v $(pwd):/myapp python:3.11.0-bullseye bash

then run the test again ...

with intel®Core i7 11800 i got these results :

Benchmark	py3.10.8	py3.11.0
NO_thread	10.16 s	8.97 s -> 12% faster
Threads(5,)	50.61 s	45.14 s -> 11% faster
Processes(5,)	15.64 s	14.61 s -> 7 % faster

The results are jittery any time you run the tests, anyway, Python 3.11 is faster .

You might modify or enhance multi-proc_thread.py to create more robust Benchmarking.